Fundus Photography Device Dirt Detection by Image Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Small amounts of dirt on fundus photography devices can significantly affect the interpretation of eye examination results, necessitating immediate detection and removal to ensure accurate readings.
Innovation Solution
An automatic dirt detection method using fundus image comparisons to identify dirt on the lens by analyzing grayscale values and contour contents of fundus images, employing a saturation channel, binarization, and object box creation to determine correlation and similarity indices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If fundus photography device is used to examine eye diseases, then examination capability is provided, but dirt on the lens affects interpretation accuracy
Solution Approach 1:
The system performs preliminary dirt detection by comparing the current fundus image with a reference fundus image before final interpretation. This preliminary action identifies dirt contamination on the lens, allowing for cleaning or re-examination before the examination results are finalized, thus preventing inaccurate interpretation caused by dirt.
2Reliability
If dirt detection is performed manually, then dirt can be identified, but inspection time is extended and efficiency is reduced
Solution Approach 1:
The system replaces manual dirt inspection with an automated image processing system. The processor automatically compares the current fundus image with a reference image, calculates correlation coefficients, and determines the presence of dirt without requiring manual intervention. This substitution of mechanical/manual inspection with an automated computational system maintains high detection accuracy while significantly reducing inspection time.
3Measurement precision
If fundus images are compared using complex algorithms, then dirt detection accuracy is improved, but device complexity increases
Solution Approach 1:
The system extracts only the essential features for dirt detection by comparing grayscale values and calculating correlation coefficients between corresponding regions of the current and reference fundus images. This extraction approach focuses on the most relevant information (intensity variations indicating dirt) while ignoring unnecessary details, thereby achieving high detection precision without requiring overly complex processing algorithms.
Data Source
AI summary
A fundus photography device, an electronic device and an automatic dirt detection method are provided. The automatic dirt detection method includes the following steps. A first fundus image is obtained. A saturation channel is used to obtain a first object box from the first fundus image. A second fundus image is obtained. Grayscale values of the first fundus image and the second fundus image corresponding the first object box are compared to obtain a first correlation coefficient. If the first correlation coefficient is greater than a correlation coefficient threshold, the fundus photography device is deemed that there is a dirt. Contour contents of the first fundus image and the second fundus image corresponding the first object box are compared to obtain a first similarity index. If the first similarity index is greater than a similarity threshold, the fundus photography device is deemed that there is a dirt.